Classification of hand movements by using artificial neural network

In this study, a home-made four channel sEMG amplifier circuit was designed for measuring EMG signals. The recorded sEMG signals were filtered with a band pass filter and afterwards wavelet based filtering was applied to remove unwanted noises. As a second step the recorded and denoised signals´ features were extracted. For classification of motions 8 time domain and 2 frequency domain features were used. There is no reduction applied to the features for artificial neural network (ANN) classification while the features were reduced in two dimension using by Diffusion Map for fuzzy classification. Lastly, seven different motions were classified by ANN and Gustafson Kessel algorithm. Also, their classification performances were compared.